Biochemistry 1983, 22, 2821-2828
2821
Turnover and Inactivation of Bacterial Citrate Lyase with 2-Fluorocitrate and 2-Hydroxycitrate Stereoisomerst Steven E. Rokita and Christopher T. Walsh*
ABSTRACT: Bacterial citrate lyase catalytically cleaves the carbon skeleton of the naturally occurring fluorocitrate isomer (-)-erythro-2-fluorocitrate (2R,3R) with the same regiospecificity as with citrate cleavage. The carbon-skeleton cleavage rate of this analogue is 1% of the V,,, for citrate turnover, and it binds to the enzyme with a KIvalue of 4.9 p M , 300-fold lower than the K,,, of citrate. Cleavage of the analogue yields oxalacetate and a fluoroacetyl form of the resting citrate lyase (vs. the acetyl form after citrate turnover). (-)-erythro-2Fluorocitrate inactivates citrate lyase ca. 500-fold more frequently than citrate (ca. 20 vs. 6500-12000 catalytic cycles per enzyme inactivation), producing the deacetylated, inactive form of the enzyme. The (+)-erythro-2-fluorocitrate (2S,3S) is also cleaved very slowly (0.05% the V,, for citrate) but does not cause measurable enzyme inactivation (over 100 min). Even when the turnover rate of citrate lyase is reduced 2000-fold as above, the regiospecificity of processing remains the same as that of citrate, yielding 0-fluorooxalacetate and the acetyl form of resting citrate lyase. Among the four diastereomers of 2-hydroxycitrate, three forms-(+)-threo, (-)-threo, and (-)-erythrcare catalytically cleaved by citrate
lyase. The regiospecificity of processing in all cases is invariant and identical with that of citrate cleavage. (-)-erythro-2Hydroxycitrate (2R,3S) and (-)-threo-2-hydroxycitrate (2S,3S) yield oxalacetate and the glycolyl form of resting citrate lyase in the presence of either zinc or magnesium ions. (+)-threo-2-Hydroxycitrate (2R,3R) yields 0-hydroxyoxalacetate and the acetyl form of citrate lyase if magnesium is present, but no catalytic activity is evident in the presence of zinc. The (+)-threo and (-)-erythro isomers inactivate citrate lyase with partition ratios of less than 100 catalytic cycles per enzyme inactivation. (+)-erythro-2-Hydroxycitrate (2S,3R) appears to be a completely efficient turnover-dependent inactivator with a khd of 0.68-1.1 min-' depending on the metal ion present. This isomer effects net hydrolysis of the active, acetyl form of the enzyme to a catalytically inactive form without producing any detectable carbon-skeleton cleavage. Enzyme inactivation induced by all of the above citrate analogues may follow a common path. The hydrolysis of a catalytic intermediate, the mixed-acid anhydride, could be the sole cause of enzyme inactivation.
F l u o r i n e has become a common substituent in pharmacologically active organic compounds [for reviews, see Peters (1972), Filler (1976), and Walsh (1983)l. A broad understanding of the enzymic reactions at or adjacent to carbonfluorine bonds is necessary for predicting the metabolism and toxicity of fluorinated substrate analogues. This paper reports the interactions between bacterial citrate lyase (EC 4.1.1.3.6) and the (+)- and (-)-erythro-2-fluorocitrateenantiomers and focuses on the stereochemistry of carbon-skeleton cleavage and turnover-dependent inactivation. Citrate was chosen as the parent compound for two major reasons. First, predictions can be made and tested concerning the metabolism of citrate derivatives because the enzymes that process citrate have been well characterized (Srere, 1975; Glusker, 1971). Second, the ability of a fluorine substituent to mimic the properties of a substrate hydrogen or hydroxyl group can be studied within a set of citrate derivatives [see also Marletta et al. (1981) and Rokita et al. (1982)l. Replacement of a methylene hydrogen of citrate with fluorine will not greatly affect the steric properties of this compound (C-H bond length is 1.09 A in methane; C-F bond length is 1.39 A in fluoromethane) but will increase the polarity of the substituted carbon [see p 95 in Peters (1972)]. The polarity of fluorine and its ability to accept hydrogen bonds more closely resemble the properties of a hydroxyl group than those of the parent hydrogen in citrate. To compare the reactivity of a fluorine vs. a hydroxyl-group substitution at C-2
of citrate, this paper also details the interactions between the 2-hydroxycitrate stereoisomers and citrate lyase. Previously, the reactivities of these hydroxylated analogues with citrate lyase were briefly surveyed by Sullivan et al. (1977a). When a substituent replaces one of the methylene hydrogens in the prochiral citrate parent, two chiral centers are formed as shown in Scheme I-one at the substituted carbon and the other at carbon 3.' The stereochemistry of these analogues defines their biological activity. Only the (-)-erythro stereoisomer of 2-fluorocitrate is toxic (Fanshier et al., 1964). This stereoisomer is produced in vivo from fluoroacetate by what Peters (1972) described as the "lethal synthesis". Two hydroxylated citrate derivatives occur in nature; (+)erythro-2-hydroxycitratecan be isolated from the hibiscus plant, and (-)-threo-2-hydroxycitrate can be isolated from garcinia fruit (Lewis & Neelakantan, 1965). Only (-)threo-2-hydroxycitrate inhibits the formation of triglycerides and cholesterol in vivo (Sullivan et al., 1977b). The activity of each 2-fluoro- and 2-hydroxycitrate stereoisomer with citrate lyase addressed here is consistent with the proposed mechanism of catalysis. This enzyme is unusual
From the Departments of Chemistry and Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 021 39. Received December 17, 1982; revised manuscript received March 3, 1983. This investigation was supported in part by National Institutes of Health Grant GM 2001 1 .
0006-2960/83/0422-2821$01.50/0
I The stereochemistry of the citrate derivatives is designated by the erythro and threo nomenclature to illustrate possible steric equivalence between 2-fluoro- or 2-hydroxycitratestereoisomers. The (-)-eryrhro2-fluorocitrate, for example, is sterically equivalent to the (-)-erythro2-hydroxycitrate. Furthermore, the numbering system for the citrate carbon skeleton used here is a historic and widely quoted one. The Cahn-Ingold-Prelog stereochemical designations are added parenthetically. Stereoequivalent fluorinated and hydroxylated citrate derivatives will not necessarily have the same R / S assignment due to the priority system for carbon-bound substituents. An alternative nomenclature for derivatives of citrate (not shown) has been developed from the parent citrate molecule (Glusker & Srere, 1973).
0 1983 American Chemical Society
2822
BIOCHEMISTRY
ROKITA A N D WALSH
Scheme I
Tow
citrate
I-) erythro
(4 erythro
2 -f Iuoroci trot e
2-f boroci trate
(2S, 3s)
(2R, 3R)
because it contains an essential acetyl group for activity (Buckel et al., 1971; Srere et al., 1972). This moiety is bound as a thiol ester to a modified coenzyme A covalently attached to one of the three subunits of citrate lyase. Another subunit of this enzyme catalyzes acyl exchange between the active site bound acetyl group and citrate (eq 1). The third subunit, in coo-
+
H O e O -
coo0
A
S-enz
the presence of certain divalent metals, then catalyzes the carbon-skeleton cleavage of citryl-S-lyase2 (eq 2), releasing ro S-enz E /E
-
'coo-
free oxalacetate and regenerating the acetyl enzyme (Dimroth & Eggerer, 1975a). Citrate lyase also autoinactivates (eq 3) in the presence of 0
A
S-enz
(active)
M++ + H20 + Citrate
Enz
-so
t CH3COO-
(3)
(inact Iv e )
citrate. This process is attributed to a metal-dependent hydrolysis during the transacylation step resulting in a net deacetylation of the enzyme (Singh & Srere, 1971). All kinetic data reported here have been measured in the presence of both Mg2+ and Zn2+ in order to facililate a comparison of our data and those previously reported in the presence of only Mg2+ or Zn2+.
* Abbreviations: MDH, malate dehydrogenase; unit, enzyme unit defined to be the quantity of enzyme necessary to transform 1 Fmol of substrate in 1 min; NAD+ and NADH, nicotinamide adenine dinucleotide, oxidized and reduced, respectively; cpm, counts per minute; HPLC, high-performance liquid chromatography; acetyl-S-lyase, the active form of citrate lyase with an acetylated, covalently bound modified coenzyme A; HS-lyase, citrate lyase (inactive) with the acetyl group removed; citryl-S-lyase, 2-hydroxycitryl-S-lyase,3-fluoro-3-deoxycitrylS-lyase, 2-fluorocitryl-S-lyase, fluoroacetyl-S-lyase, and glycolyl-S-lyase, citrate lyase with the appropriate acyl moiety attached through a thiol ester tothe coenzyme A-like group..
(+I erythro 2-hydroxycitrate (2S, 3R)
(4threo 2-hydroxycitrate (2S,3S)
Materials and Methods The synthesis of /3-fluorooxalacetate has been described by Goldstein et al. (1978). The (+)- and (-)-erythro-2-fluorocitrate enantiomers were synthesized and resolved according to known procedures (Dummel & Kun, 1969). [Caution: handle (-)-erythro-2-fluorocitrate with extreme care. It is highly toxic.] The resolved (-)- and (+)-threo-2-hydroxycitrate and (-)- and (+)-erythro-2-hydroxycitrate stereoisomers were the gift of Dr. Ann Sullivan from the Department of Biochemical Nutrition of Hoffmann-La Roche Inc. (Nutley, NJ). (4R)-[4-3H]NADH (1.7 pCi/pmol) was the gift of Dr. C. Ryerson from this laboratory and was prepared according to the procedures of Oppenheimer et al. (1971). All other chemicals and reagents were of the highest quality commericially available. Citrate lyase from Klebsiella aerogenes was purchased from Sigma Chemical Co. in a preparation containing, by weight, 24% bovine serum albumin, 48% saccharose, 24% citrate lyase (6.1 units/mg), and 4% MgS04.7H20. This mixture was used without purification of citrate lyase because this enzyme is less stable when the other components are removed. This enzyme was assayed by coupling the formation of oxalacetate (or its fluoro or hydroxy derivatives) to its subsequent reduction with NADH and MDH (Bergmeyer, 1974). All kinetic assays were performed at 25 OC and buffered with 100 mM triethanolamine at pH 7.6. Between 48 and 240 pg of enzyme protein (from the citrate lyase preparation) was used for k,,, determinations; the concentrations of the citrate derivatives were at least 10-fold greater than their KI [this paper and Sullivan et al. (1977a)l to approximate maximum rate conditions. Turnover inactivation was never too fast to obscure the initial rate of carbon-skeleton cleavage over the first 2 min. The concentration of metal ions present in the assays (2 mM) exceeded, in every case, the concentrations of the carbon substrate. Because Mg2+was present in the Sigma preparation of citrate lyase, assays containing zinc were contaminated with magnesium. A K I value for (+)-erythro-2-fluorocitratewas calculated from its competitive inhibition of citrate cleavage (determined by a Lineweaver-Burk plot). The KI and kinact values for the time-dependent inactivation of citrate lyase by (-)-erythro2-fluorocitrate were determined as described for 3-fluoro-3deoxycitrate (Rokita et al., 1982). The specific activity of citrate lyase (96 pg of enzyme protein) incubations containing varying concentrations of inhibitor at 25 OC in 200 p L of 1 and 2 m~ zn2+-100 m~ triethanolamine, p~ 7.6, was monitored Over time. A KI and kmact were based On a double-reciprocal plot of the observed k,,,,, VS.the inhibitor concentration. Values of k4nnp, for (+)-three- and (+)-erythro2-hydroxycitrate were from the decrease in the specific activity of citrate lyase (240 pg) when incubated in a 2oo-ML solution of 2 mM Zn2+-80 mM triethanolamine, pH 7.6, and inhibitor concentrations 10 times their published .
I
I
VOL. 2 2 , NO. 1 2 , 1983
CITRATE LYASE A N D SUBSTRATE ANALOGUES
KI (Sullivan et al., 1977a). The kinaclvalue for (+)-erythro2-hydroxycitrate in the presence of 2 m M Mg2+ was too large to measure as above; instead, a kinactwas extrapolated as in the case of the (-)-erythro-fluoro derivative. The turnover inactivation constants, kinact,and partition ratio (catalytic turnovers per enzyme inactivation), were calculated for citrate and its analogues from the initial rate of carbon-skeleton cleavage and the total number of moles of substrate cleaved per mole of catalytically competent active site (see legend of Table 11). Although native citrate lyase (M,of 550000) contains six subunit complexes (Bowen & Mortimer, 1971), the above calculations assume that only four of these complexes are functional (Dimroth & Eggerer, 1975b; Singh et al., 1975). The addition of 2-3 pL of reagent-grade acetic anhydride to a routine activity assay was sufficient for the acetic anhydride reactivation of citrate lyase (Buckel et al., 1971; Srere et al., 1972; Rokita et al., 1982). The a-keto acid cleavage products of citrate and its derivatives were reduced in situ by (4R)-[4-3H]NADH and MDH. A 600-pL enzyme incubation of 0.3-0.8 m M ZnC12, 70 p M [3H]NADH, 3 mM citrate analogue, 80 mM triethanolamine, pH 7.6, 1.2 mg of citrate lyase, and 100 units of M D H was quenched after 2-6 h by the addition of HCl to a final pH of 3-4. Before the mixture was separated by anion-exchange HPLC (see Figure l), the nucleotides were removed from this incubation with activated charcoal, followed by filtration through a 0.45-pm Millex-HA aqueous filter (Millipore Corporation), and 1 pmol of malate and 3 pmol of tartrate were added as nonradioactive standards. Results and Discussion Citrate Lyase Catalyzed Cleavage of 2-Fluoro- and 2Hydroxycitrate Stereoisomers: Product Identification. Before the chemistry of the 2-fluoro- and 2-hydroxycitrate stereoisomers can be discussed, their catalytically competent alignment in the active site of citrate lyase must be defined. Carrel1 et al. (1970), Sullivan et al. (1977a), and Stallings et al. (1979) have suggested that a fluoro- or hydroxyl-group substitution at the methylene position of citrate may cause these derivatives to bind citrate-processing enzymes in a manner not analogous to the binding of citrate. However, the regiospecificity of citrate synthase (EC 4.1.3.7) and the mammalian ATP citrate lyase (EC 4.1.3.8) catalysis is constant for citrate [for a review, see Srere (1975)l and the 2fluorocitrate stereoisomers (Fanshier et al., 1964; Marletta et al., 1981). These enzymes catalyze a reversible aldol-type condensation in which the carbon bond cleaved or produced always occurs at the arm corresponding to the pro-S arm of citrate. Rokita et al. (1982) reported a MDH-coupled citrate lyase activity for both of the erythro-2-fluorocitrates,and Sullivan et al. (1977a) reported that three of the four 2-hydroxycitrate stereoisomers were also substrates using the same MDHcoupled assay. However, M D H will catalyze the reduction of all of the possible a-keto acid products, oxalacetate, pfluorooxalacetate (Marletta et al., 198 l ) , and p-hydroxyoxalacetate (Sullivan et al., 1977a). The stereochemistry of citrate lyase turnover and, therefore, the orientation of productive binding can only be determined in each case after the cleavage products are identified. 0-Fluorooxalacetate, produced by citrate lyase, is sufficiently stable to be detected by HPLC without prior derivatization. The other possible four-carbon keto acid fragments were successfully identified by HPLC only after in situ reduction with (4R)-[4-3H]NADH and M D H (see Figure 1). Figure 2 summarizes the cleavage products identified after
3000
2823
i
A
:
E
75t
8
16
24 min
32
40
FIGURE 1: HPLC separation of citrate lyase processed (+)-threo-2hydroxycitrate after reduction with [3H]NADH. The citrate lyase incubation (described under Materials and Methods) was analyzed on a Waters Associates HPLC (including a 660 gradient programmer) equipped with a Micromeritics 786 variable-wavelength detector and an analytic Whatman Partisil SAX strong anion exchange column (protected by a CI8guard column). (A) HPLC fractions (2 mL) were mixed with 15 mL of scintillant and analyzed for tritium content on a Beckman LS-100 scintillation counter. Note: there is ca. a 3-mL dead volume between the variable-wavelength detector flow cell and the fraction collector. (B) Nonradioactive malate and tartrate carrier were added to the enzyme incubations and detected by their absorbance at 210 nm. (C) The HPLC base line rose in parallel with the indicated gradient of potassium phosphate, pH 3.8, with a flow rate of 2 mL/min.
citrate lyase turnover of the 2-fluoro- and 2-hydroxycitrates. Similar to citrate synthase and ATP citrate lyase (Fanshier et al., 1964; Marletta et al., 1981), citrate lyase processes the fluorinated substrates with the same regiospecificity as citrate. If cleavage with the reverse regiospecificity competes during catalysis, it must occur less than 2% of the time. The cleavage of (-)-erythro-2-fluorocitrate (2R,3R) yields oxalacetate, suggesting that the enzyme must be left at the end of the first catalytic cycle in the fluoroacetyl form (fluoroacetyl-S-lyase) instead of the acetyl form (acetyl-S-lyase). Citrate lyase remains active in this form because the modified enzyme initiates further catalytic cycles. In the 2-hydroxycitrate series (Figures 1 and 2), the only productive binding for these analogues is one that mimics that of citrate [confirming the predictions of Sullivan et al. (1977a)l. For the three stereoisomers that are substrates for citrate lyase, cleavage always occurs at what corresponds to the pro-S arm of citrate. Just as the catalytic turnover of (-)-erythro-2-fluorocitrate demonstrates that fluoroacetate can substitute for the active site acetyl group, the catalytic turnover of (2R,3S)-(-)-erythro- and (2R,3R)-(-)-threo-2-hydroxycitrate demonstrates that a glycolyl form of citrate lyase (glycolyl-S-lyase) is also catalytically active. Only one alignment of the six-carbon substrate will undergo catalytic cleavage despite the variety of ways citrate and its derivatives may chelate divalent metal ions (Stallings et al., 1979, 1980; Glusker, 1980). Neither a fluoro nor a hydroxy
2824
BIOCHEMISTRY
ROKITA AND WALSH
CWH
l
Table I: Carbon-Skeleton Cleavage Rates of Citrate Derivatives
4bCKOOH
Catalyzed by Citrate Lyase' turnover
turnover
(kcat) with
(kcat) with 2 mM Mg2+
2 mM Znz+ compd
citrate (-l-erythro-2-
(min-')
(min-')
750 6.3
1100 2.6
fluorocitrate ( t )-eryrhro-2-
fluorocitrate (-)-erythro-2hydroxycitrate (+)-eryrhro-2-
0.38
0.63
1.2
1.0 d HOOC
threo
C-OH
HOOC
P
P h y d r o a y n t r a t e 4 c,,r,,~
d.M
'n-tortroC
reductlon
lyiie
It1ermro ~ f b J o r 0 c l t m f r 6-n
d-F
0.45
hydroxycitrate
cnrota Iyoee
I*)
0.35
(-)-threo-2-
cn3coon
f Iuorooiolocatatc
FIGURE 2: Citrate lyase cleavage pattern with citrate analogues. The regiospecificity of citrate lyase processing is summarized here by identifying the products of catalytic turnover either directly or after reduction with ( 4 R ) - [ 4 J H ] N A D Hand MDH. Enyzme incubations are described under Materials and Methods and were analyzed by HPLC as illustrated in Figure 1. The metabolite that contained the tritium is listed on the right. @-Fluorooxalacetatewas identified directly by coelution with a standard solution of @-fluorooxalacetate(using the same HPLC columns but an isocratic elutant of 100 mM potassium phosphate, pH 3.8).
substituent realigns the substrate-enzyme binding geometry to allow for any loss of stereospecific processing, even though the rate of catalysis may be decreased more than a 1000-fold by the added substituent (see below). Binding Recognition and Catalytic Carbon-Bond Cleavage of the 2-Fluorocitrates with Citrate Lyase. The chemical effects of the fluorine substituent in 2-fluorocitrate during substrate binding and turnover can now be divided into the effects of a fluorine geminal, (-)-erythro-2-fluorocitrate (2R,3R), and vicinal, (+)-erythro-2-fluorocitrate (2S,3S), to the carbon bond cleaved (Figure 2). Because these analogues are processed very slowly (see below), active site binding was initially assessed by measuring competitive inhibition of citrate turnover. The KI for (+)-erythro-2-fluorocitrateis 86 pM, almost 20-fold lower than the K , found experimentally for citrate ( K , = 1.6 mM). A KI from competitive inhibition assays could not be determined for (-)-erythro-2-fluorocitrate because it quickly and irreversibly inactivated citrate lyase. Instead, a KI of 4.9 p M for the (-)-erythro isomer was derived from time-dependent inactivation studies. This is almost 3 orders of magnitude lower than the citrate K,, The k,, values for 2-fluorocitrate turnover, listed in Table I, are only 0.05% [(+)-erythro isomer] to 1% [(-)-erythro isomer] the k,, for citrate. (-)-erythro-2-Fluorocitrate (with the fluorine geminal to the bond broken) is cleaved 20-fold faster than (+)-erythro-2-fluorocitrate (with the vicinal fluorine). However, mammalian ATP citrate lyase exhibits the exact opposite order of reactivity with these stereoisomers (Marletta et al., 1981). Clearly, the fluorine substituent at either position does not have the same effect during turnover of these enzymes. Those characteristics of a fluorine substitution (e.g., inductive effect) that remain constant during the turnover of both enzymes cannot then explain the relative
